Rate the end-to-end reliability of the tools on multi-step tasks, from first step to finished output

The mean rating was 4.5 and the median 5.0 (n = 782 respondents). Ratings of 3 or below, the distressed range, made up 32.48% of the answers. The rating was associated with all four study measures. Higher ratings came with less cancellation (r = −0.65, 95% CI −0.74 to −0.55, p < 0.001, n = 150 engagements), higher customer satisfaction among active engagements (r = 0.65, 95% CI 0.52 to 0.75, p < 0.001, n = 94 engagements), a higher engagement indicator mean computed without the indicator (r = 0.85, 95% CI 0.80 to 0.89, p < 0.001, n = 150 engagements) and lower intent to leave (r = −0.65, cluster bootstrap 95% CI −0.71 to −0.59, p < 0.001 with standard errors clustered by engagement, n = 741 respondents). After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 44.95, Holm p < 0.001, omega squared 0.415), respondent role (Welch F 7.55, Holm p = 0.006, omega squared 0.016) and vendor type (Welch F 4.27, Holm p = 0.006, omega squared 0.021).

Of the 51 written responses, nearly all from C177, the comment field on the AI tools, 36 were negative and 13 positive. Respondents described tools that handled single steps well and lost a step or the thread on long chains, then reported the run complete. Teams responded by cutting work into single steps with a person between them. A more negative written score came with higher intent to leave.

Ratings of 3 or below, the distressed range, accounted for 254 of the 782 answers (32.48%). The engagement mean fell in the distressed range in 36 of 150 engagements.

The primary responses numbered 49, all written in answer to C177, “Add any comments on the agentic AI tools that the ratings in this section do not capture.” The remaining 2 responses, 3.92% of the total, were secondary and came from 2 other questions. That secondary share ranked 53rd from the highest among the 71 Part 2 indicators with 30 or more written responses, against a median of 18.92%. Respondents who wrote both a primary and a secondary response numbered 0. Negative responses made up 69.39% of primary responses (34 of 49) and 2 of the 2 secondary responses.

Twenty-one responses described the tool losing a step, a rule or the thread partway through a long task: a six-step attribution run that dropped the currency effect at the end, a deploy job of eleven steps that did something odd on the tenth in about one run in ten, a covenant dropped between spreading and the early-warning note, a fund dropped from an allocation near the end. Several respondents said the tool was near perfect on a two-step task and unusable on a six-step one, and that one rating could not hold both.

Twelve responses described runs reported complete or passed when steps had failed. A recertification run reported four hundred accounts complete when about thirty entitlements had not been revoked. A batch of eighty investigations taken to the financial institution's financial crime committee showed all complete when six had stopped part way, and the respondent said that conversation set the programme back a month. The single response to C055 described failed settlement runs excluded from the reported success rate. Five responses described cutting work into single steps or checkpointed pieces with a person signing off between them. In a bad week this was slower than doing the work by hand.

The thirteen positive responses described tools that finished long runs or whole pipeline jobs, often as an improvement on a previous engagement, and most named a remaining weakness in sources, logging, scope or consistency.

Of the 51 responses, 36 (70.59%, 95% CI 56.25% to 83.64%) were negative and 4 (7.84%) carried the strongest negative score of −10. Positive responses numbered 13 (25.49%), and 2 were neutral. The mean written score was −2.4 (95% CI −3.9 to −0.7, n = 51 responses), and the median was −5.0.

Of the 51 responses, 50 came from respondents who also rated the indicator. Among them, the written score rose with the rating (Spearman rho 0.65, 95% CI 0.47 to 0.79, p < 0.001, n = 50 responses). The written responses corroborated the rating. Of the 35 negative responses with a rating, 14 (40.00%) came from respondents whose rating was in the distressed range.

Respondents who wrote about the indicator and gave a rating numbered 50, with a mean rating of 4.7. The 732 respondents who rated the indicator without writing about it had a mean of 4.5. The difference, 0.1 (95% CI −0.5 to 0.8), included zero. Ratings in the distressed range made up 28.00% of the ratings from respondents who wrote (14 of 50) and 32.79% of the ratings from respondents who did not (240 of 732).

Every figure in the segment tables is at respondent level. Segments with fewer than 30 respondents carry no interval. Segments with fewer than 10 report counts only.

Segment families recorded at engagement level were tested on engagement means. Respondent role and vendor type were tested on respondents. Where an expected count fell below 5, the test of the distressed share used Fisher's exact test for a two-by-two table, marked exact, or a Monte Carlo test, marked simulated. After Holm correction, ratings differed by engagement status and lifecycle stage (Welch F 44.95, Holm p < 0.001, omega squared 0.415). Active engagements in Build-Out rated highest (mean 5.4, n = 167 respondents) and canceled engagements lowest (mean 3.2, n = 297 respondents). Ratings also differed by respondent role (Welch F 7.55, Holm p = 0.006, omega squared 0.016). Portfolio directors rated highest (mean 5.3, n = 29 respondents) and individual contributors lowest (mean 4.4, n = 602 respondents). Ratings also differed by vendor type (Welch F 4.27, Holm p = 0.006, omega squared 0.021). Consulting firms rated highest (mean 5.2, n = 105 respondents) and pure-play agentic AI firms lowest (mean 4.1, n = 69 respondents). Outsourcing model, institution type, process family and class of work showed no difference.

Written sentiment did not differ by engagement status and lifecycle stage across the 46 engagements with a written response (Kruskal-Wallis H 1.26, df 3, p = 0.738). Of the groups with 10 or more responses, responses from canceled engagements were the most negative (mean −3.1, 13 responses) and responses from active engagements in Build-Out the least negative (mean −1.5, 10 responses). Responses from active engagements in Production (7) are reported as counts only.

No test across respondent roles was run, because fewer than two roles had 30 or more written responses. Individual contributors wrote 33, engagement managers wrote 17 and portfolio directors wrote 1 responses.

Cancellation rate is the share of engagements canceled by the study date, measured at engagement level (150 engagements, 56 canceled). Canceled engagements were rated after cancellation. The association is therefore a descriptive contrast and not a forecast of cancellation.

Across the study set, higher ratings came with less cancellation (r = −0.65, 95% CI −0.74 to −0.55, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in fourteen of the 20 segment levels tested.

Customer satisfaction is the percentage recorded against each engagement. The primary estimate uses the 94 active engagements. A descriptive contrast uses all 150 engagements.

Across the active engagements, higher ratings came with higher customer satisfaction (r = 0.65, 95% CI 0.52 to 0.75, Holm p < 0.001, n = 94 engagements). The association held after Holm correction in eleven of the 16 segment levels tested.

The engagement indicator mean is the mean of an engagement's ratings across the rated indicators, measured at engagement level (150 engagements). The indicator is one of the rated indicators. Each comparison therefore uses the mean recomputed without it.

Higher ratings came with a higher engagement indicator mean (r = 0.85, 95% CI 0.80 to 0.89, Holm p < 0.001, n = 150 engagements). The association held after Holm correction in twenty-four of the 24 segment levels tested.

Intent to leave is the answer to C019, “How likely are you to leave your current job in the next 12 months?”, rated 1 to 10. It is measured at respondent level on up to 798 respondents, with the engagement-level form reported alongside.

Respondents who gave higher ratings reported lower intent to leave (r = −0.65, 95% CI −0.69 to −0.61, Holm p < 0.001, n = 741 respondents). The association held after Holm correction in thirty-six of the 37 segment levels tested.

Two comparisons set the engagements with at least one negative response against all other engagements. Of the 118 other engagements, 104 had no written response addressing the indicator. These two comparisons therefore do not isolate written sentiment. Of the 32 engagements with a negative response, 10 (31.25%) were canceled, against 46 of the 118 other engagements (38.98%). The difference did not reach significance (Fisher exact p = 0.537). Customer satisfaction among active engagements did not differ between the two groups (mean 62.18% on 22 engagements against 64.68% on 72, Welch t −0.81, p = 0.426).

Two correlations test written sentiment directly. Across the 46 engagements with a written response, the engagement mean written score was not associated with cancellation (point-biserial r −0.18, p = 0.238). Across 49 respondents, a more negative respondent mean written score came with higher intent to leave (Spearman rho −0.35, p = 0.015). Of the four comparisons, one reached significance at 0.05.

The four associations were compared in pairs on their engagement-level forms, each oriented so that a higher value is the better outcome, across the 150 engagements with all four measures. After Holm correction, five of the six pairs differed. The association with the engagement indicator mean was the strongest and the association with customer satisfaction the weakest. The associations with customer satisfaction and intent to leave did not differ (Holm p = 0.996).

Answers clustered within engagements (ICC 0.58). Clustering reduced the 782 respondent answers to the equivalent of about 227 independent answers. Respondent-level intervals therefore used the cluster bootstrap. Respondent-level tests of intent to leave used standard errors that allow for clustering by engagement.

No association was equivalent to zero. Each of the four differed from zero. The detection floor shown is the study floor at 150 engagements. Intent to leave was tested on its effective sample size.

Each change was computed before rounding and may differ by 0.01 from the difference between the two printed values.

All 22 checks held. A check holds when its test falls on the same side of 0.05 as the primary test. Where the primary test falls below 0.05, the check must also keep the sign of the primary correlation. The thin-base check removes the engagements in any level of outsourcing model or process family with fewer than 30 engagements. Institution type is not used for that check, because every one of its levels has fewer than 30 engagements.

Each of the four measures was computed from data the indicator does not contain. Each result is therefore a finding and not an arithmetic consequence. The indicator is one of the rated indicators, and the engagement indicator mean was recomputed without it. The correction changed r by 0.00, from 0.85 to 0.85.

Each deviation was computed before rounding and may differ by 0.01 from the difference between the two printed values.

The comparison covered 150 engagements. The indicator followed the study's typical pattern on all four measures, and no deviation held after Holm correction.

Among the 71 Part 2 indicators with 30 or more written responses, the indicator's share of negative responses, 70.59%, ranked 59th from the highest, against a median of 80.82%. Its mean written score, −2.4, ranked 66th from the most negative, against a median of −4.7.

The indicator most often present in the same responses was C172, “Rate how accurately the tools' own reports of success or failure matched the actual outcomes of their work.”, in 16 of the 51 responses (31.37%). C157, “Rate the factual reliability of the AI tools' outputs on this engagement before any human review.” (11 responses), C176, “Rate the quality of the tools' logs, traces, and usage telemetry for detecting and diagnosing failures.” (8 responses), C165, “Rate how completely the tools captured material facts located anywhere in long documents, including mid-document content.” (7 responses) and C168, “Rate how consistently the agents took only intended and authorized actions when operating connected tools and systems.” (7 responses) followed.

The most frequent theme was “Comparison to a prior engagement”, in 8 responses (15.69%). The second was “The scale or question does not fit”, in 6 responses.

About a third of all ratings of the end-to-end reliability of the tools on multi-step tasks fell in the distressed range (254 of 782 answers, 32.48%). The rating was associated with all four study measures, and every association held under each of the 22 robustness checks. Canceled engagements rated far below the three active stages (mean 3.2, n = 297 respondents).

The written evidence describes reliability that falls with the length of the chain, combined with a completion report that does not fall with it. The second is the more costly, because it removes the signal that a person needs to check. Written sentiment was associated with intent to leave.

The evidence is the reported experience of the vendor's delivery teams, collected in a single cross-section. The associations reported here are not causal.

Tests by respondent role and by vendor type treated respondents as independent. The section on clustering reports how far clustering within engagements reduced the information in respondent-level figures.

Customer management executives gave 1 answers, a base of anecdote strength.

Shares of written sentiment are shares of responses and not of the study set of respondents. The written score takes five fixed values, so its mean and median are coarse. Responses clustered within engagements (51 responses on 46 engagements). Intervals for written sentiment therefore used a bootstrap by engagement.